CATALYTIC PYROLYSIS OF POLYOLEFINS

Transcription

1 CATALYTIC PYROLYSIS OF POLYOLEFINS A Thesis Presented to The Academic Faculty by Ifedinma Ofoma In Partial Fulfillment of the Requirements for the Degree Masters of Science in Chemical & Biomolecular Engineering Georgia Institute of Technology May 2006

2 CATALYTIC PYROLYSIS OF POLYOLEFINS Approved by: Dr. John Muzzy, Advisor School of Chemical & Bimolecular Engineering Georgia Institute of Technology Dr. Christopher Jones School of Chemical & Bimolecular Engineering Georgia Institute of Technology Dr. Matthew Realff School of Chemical & Bimolecular Engineering Georgia Institute of Technology Date Approved: 11 th of January, 2006

3 ACKNOWLEDGEMENTS I wish to express my appreciation to my advisor, Dr. John Muzzy for his guidance and assistance during my studies and research endeavor at Georgia Tech. Thank you for your patience and understanding. I must also acknowledge my committee members, Dr. Christopher Jones and Dr. Matthew Realff for their patience, valuable contributions and resources towards my thesis project. I am also very grateful to Latoya Bryson for assisting me along the way through my time here. Many people assisted me in the various experimental procedures; to them I offer my deepest appreciation: Dr. Beckham s group for their assistance with the thermogravimetry (TG) equipment, especially Chris Hubbell for training me in TG analysis. I would also like to thank John Richardson for assisting me in my catalyst characterization and synthesis. iii

4 TABLE OF CONTENTS Page ACKNOWLEDGEMENTS LIST OF TABLES LIST OF FIGURES SUMMARY iii vii x xvi CHAPTER 1 INTRODUCTION 1 2 LITERATURE REVIEW Thermal Pyrolysis of Polyolefins The Hydrocracking Mechanism Thermal Pyrolysis Product Yields Catalytic Cracking of Polyolefins Catalytic cracking pathway Zeolites in polyolefin pyrolysis ZSM-5 in Polyolefin Pyrolysis FCC catalysts in Polyolefin Pyrolysis The composition of an FCC catalyst Performance of FCC catalysts in polyolefin pyrolysis Equilibrium FCC Catalyst (E-CAT) Coke Formation in FCC Catalysts Effect of Polymer Type on Product distribution Effect of Particle/Crystallite Size on Product Distribution Process Design Catalyst Contact Mode Reactor Type Batch and Semi-Batch Reactors Fixed Bed Semi- Batch reactor Fluidized bed batch reactors Continuous Flow Reactors (CFRs) Effect of Feed Composition Effect of Other Process Parameters Kinetic Studies in Polyolefin Pyrolysis Kinetic Models Based on Thermogravimetry (TG) Isothermal TG measurement Dynamic TG measurement DTG Peak property method (DTG-PPM) Modulated Thermogravimetry (MTG) Factors affecting kinetic parameters estimated by TG Kinetic Models Based on Mechanistic Modeling 38 iv

5 Kinetic Models based on Reaction Mechanisms and Elementary reactions Kinetic Models based on Multi-step Reactions 38 3 EXPERIMENTAL METHODS Polymer Materials Composition of Polypropylene Post Consumer Carpet (PP-PCC) Preparation of PP-PCC Pellets Catalyst Materials ZSM Xray Diffraction Solid State Magic Angle Spinning (MAS) NMR Scanning Electron Microscopy (SEM) Nitrogen Adsorption Isotherms Fluid Catalytic Cracking (FCC) Catalysts Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray (EDX) Particle Size Distribution (PSD) by SEM Calcium Carbonate (CaCO3) Thermogravimetry TG Sample Preparation TG Analysis Catalyst(s) weight loss Determination of Kinetic Parameters by TG Isoconversion Method Arrhenius Equation 67 4 RESULTS AND DISCUSSION: PERFORMANCE OF CATALYSTS Catalytic degradation of Polypropylene and Polystyrene with ZSM Evaluation of FCC Catalysts in PP Pyrolysis Enhancing A-ECAT Performance in PP Pyrolysis Evaluation of FCC Catalysts in PE Pyrolysis Performance of FCC Catalysts and CaCO3 in PP-PCC Degradation Estimation of Rate Equations for PP and PE Degradation Rate Equation for PP Degradation Estimating Ea and A by the Arrhenius Equation Estimating Ea and A by the Isoconversion method Simulation of weight loss plots Rate Equation for PE Degradation Effect of Catalyst Contact Mode in TG Analysis of Polypropylene Degradation Dry Mixing: Performance of FCC catalysts Dry Mixing and Melt Mixing: Comparison of TG analysis REACTOR MODEL The Reactor Critical Assumptions in Process Modeling Reactor Design Rate Equation 123 v

6 5.2.3 Heat and Mass Transfer Limitations Reaction and Product Distribution Extruder Geometry Estimation of the Heat of Reaction Development of the Reactor Model Mass and Energy Balances Calculation of E, Energy input from extruder drive MATLAB simulation Economic Analysis Base Extruder Costs Raw materials Utility Costs Electricity Cooling Water Costs Labor Costs Results and Discussion Results: Run at 250 RPM Results: Run at 500 RPM CONCLUSIONS AND RECOMMENDATIONS 151 APPENDICES 154 APPENDIX A 155 APPENDIX B 159 APPENDIX C 172 REFERENCES 188 vi

7 LIST OF TABLES Table 2.1 Table 2.2 Table 2.3 Thermal versus catalytic pyrolysis of HDPE in a Batch reactor for 1 hr 10 Product distribution for degradation of HDPE in a fluidized bed reactor using fresh and equilibrium FCC catalyst at 360 C (C/P loading = 2:1, reaction time = 30 min) 20 Waste versus virgin pyrolysis of HDPE using ZSM-5 and under similar operating conditions. 27 Table 2-4 Influence of certain process conditions in polyolefin pyrolysis 29 Table 2.5 Literature values of the activation energy (E a ), reaction order (n) and pre-exponential (A) for degradation of PE, PP, and PS 36 Table 3.1 Nitrogen adsorption isotherm report on ZSM-5 sample 49 Table 3.2 Physical properties of Albemarle catalysts as reported by supplier 50 Table 3.3 TG sample preparation methods 57 Table 3.4 Observed weight loss in catalysts prior to and after 100 C 61 Table 4.1 Performance of FCC catalysts at 1% and 99% conversion of PP 75 Table 4.2 Temperature and conversion at maximum rate of degradation of PP 77 Table 4.3 Linear correlation of maximum degradation temperature with catalyst weight ratio 78 Table 4.4 Performance of FCC catalysts at onset and end of PE degradation. 83 Table 4.5 Temperature and conversion at maximum rate of degradation of PE 84 Table 4.6 Table 4.7 Table 4.8 Linear correlation of maximum degradation temperature with catalyst weight fraction 86 Performance of FCC catalysts and CaCO 3 on the onset of PP PCC degradation. 88 Average kinetic parameters for PP degradation using 8 wt % of FCC catalysts. 91 Table 4.9 Rate constant ratios at temperatures between 200 and 450 C 94 vii

8 Table 4.10 Table 4.11 Average kinetic parameters for PP degradation using varying weight fraction of S-ECAT 95 Apparent kinetic parameters estimated by the isoconversion method for the catalytic pyrolysis of PP using 8 wt % FCC Catalysts 99 Table 4.12 Rate constant ratios between 200 and 450 C 100 Table 4.13 Table 4.14 Apparent kinetic parameters for variation of S-ECAT in PP sample by the isoconversion method 101 Estimation of kinetic parameters for the catalytic pyrolysis of PE at 8 wt % catalyst using the Arrhenius equation within a conversion range of 4-40% 110 Table 4.15 Rate constant ratios at temperatures between 200 and 450 C 111 Table 4.16 Table 4.17 Comparing the repeatability of TGA properties between the melt mix and dry mix method of TG sample preparation at 97 wt% of PP 118 Comparing the repeatability of TGA properties between the melt mix and dry mix method of TG sample preparation at 8 wt% of catalyst 119 Table 5.1 Estimated heat of reactions for selected samples 129 Table 5.2 A comparison of H m estimated by DSC and DTA 130 Table 5.3 Rate law parameters for pyrolysis reaction 140 Table 5.4 Reaction zone specifications 141 Table 5.5 Modeling results for the catalytic pyrolysis of PP at 250 RPM 142 Table 5.6 Estimated costs for the catalytic pyrolysis of PP at 500 RPM 147 Table 5.7 Table A.1 Estimated costs for the catalytic pyrolysis of PP using multiple extruders at 500RPM 149 Recent lab-scale experiments on polyolefin pyrolysis using other reactor set ups. 155 Table A.2 Product distribution of polyethylene in fluidized bed reactors 156 Table A.3 Recent lab-scale experiments on polyolefin pyrolysis using CFRs viii

9 and modified CFRs/CFR combos. 157 Table A.4 Repeatability of TGA runs for PE 158 Table B.1 Estimation of kinetic parameters for PP degradation using 8wt % of FCC catalysts. 162 Table B.2 Estimation of kinetic parameters for PP degradation using 8wt % of FCC catalysts. 163 Table B.3 Estimation of kinetic parameters for PE degradation using 8 wt % of FCC catalysts. 171 Table C.1 Physical Properties of Apparent Gaseous Products. 172 Table C.2a Physical Properties of Apparent Liquid Products (liquid) 172 Table C.2b Physical Properties of Apparent Liquid Products (gas) 173 ix

10 LIST OF FIGURES Figure 1.1 Municipal Solid Waste Generation, Recycling, and Disposal in the United States 1 Figure 2.1 Random chain scission in polyethylene 6 Figure 2.2 Random chain scission in polypropylene 7 Figure 2.3 Unzipping mechanism in polystyrene 8 Figure 2.4 Conversion obtained in the thermal and catalytic cracking of HDPE, LDPE and PP (400 C, 0.5 h, plastic/catalyst=50 w/w). Catalysts are Al-Beta and Al-Ti-Beta zeolites 10 Figure 2.5 Structures of some common zeolites used in polyolefin pyrolysis 13 Figure 2-6 Equilibrium FCC sample containing high levels of coke (Ni: 2600; Va: 6700; Fe:7000) ppm. Substantial amount of Fe deposits were found within the white rectangle. Magnification: 750X 21 Figure 3.1 Structure of tufted carpet 42 Figure 3.2 Composition of post consumer carpet 42 Figure 3.3 XRD patterns of synthesized ZSM-5 44 Figure Si (a) and 27 Al (b) solid state MAS NMR spectrum of calcined ZSM-5 samples. 46 Figure 3.5 SEM images of ZSM-5 at 1000X 47 Figure 3.6 EDX bulk plot for synthesized ZSM-5 48 Figure 3.7 (a) SEM image of S-ECAT taken at 200X 51 (b) SEM image of Fresh FCC Fines taken at 500X 51 (c) SEM image of Fresh CAT (regular) taken at 500 X 52 Figure 3.8 EDX bulk plot for S-ECAT 53 Figure 3.9 PSD analyses by SEM for Fresh FCC catalysts 54 Figure 3.10 Figure 3.11 Schematic of mixer cross-section showing sigma blade placement in the mixing zone 56 TG plots of dry and melt mixed virgin PP (a) and PE (b). The dry run was repeated (1, 2) 58 x

11 Figure 3.12 TG Plot of FCC catalysts and CaCO 3 60 Figure 3.13 TG Plot of Fresh Catalysts 61 Figure 4.1 TG plot of PP degradation using 3 and 8 wt% of ZSM-5 and S-ECAT catalysts 71 Figure 4.2 TG plot of PS degradation using 3 and 8 wt% ZSM-5 and S-ECAT 72 Figure 4.3 (a) TG analyses of FCC catalysts in the PP pyrolysis at 97 wt% of polymer 73 (b) TG analyses of FCC catalysts in the PP pyrolysis at 92 wt% of polymer 73 (c) TG analyses of FCC catalysts in the PP pyrolysis at 87 wt% of polymer 74 Figure 4.4 Effect of catalyst weight fraction on T max in PP degradation 78 Figure 4.5 Figure 4.6 Degradation of 97 wt % PP with 3 wt % mixed catalyst where 0 wt % is all Fresh Fines and 100 % is all A-ECAT 80 (a) TG analyses of FCC catalysts in the PE pyrolysis at 97 wt% of polymer 81 (b) TG analyses of FCC catalysts in the PE pyrolysis at 92 wt% of polymer 81 (c) TG analyses of FCC catalysts in the PE pyrolysis at 87 wt% of polymer 82 Figure 4.7 Effect of catalyst weight fraction on T max in PE degradation 85 Figure 4.8 (a) TG plot of FCC catalysts and CaCO 3 in the PP-PCC pyrolysis at 97 wt% of PP- PCC 86 (b) TG plot of FCC catalysts and CaCO 3 in PP-PCC pyrolysis at 92 wt% of PP-PCC 87 Figure 4.9 Arrhenius plot for the degradation of virgin PP 90 Figure 4.10 Figure 4.11 Figure 4.12 Arrhenius plot for the catalytic degradation of PP using 8 wt% Fresh Fines 90 Plot of the rate constant, k versus temperature for the catalytic pyrolysis of PP by FCC catalysts. The kinetic parameters were derived by the Arrhenius method. 93 A plot of the rate constant, k versus temperature for PP pyrolysis using S-ECAT. The kinetic parameters were derived by the Arrhenius method 96 xi

12 Figure 4.13 Isoconversion TG data analysis method. Plot of log β versus T -1 for Virgin PP. 97 Figure 4.14 Isoconversion TG data analysis method. Plot of log β versus T -1 for 8 wt % Fresh fines. 98 Figure 4.15 Figure 4.16 Figure 4.17 Figure 4.18 Figure 4.19 Figure 4.20 Figure 4.21 Figure 4.22 Figure 4.23 Plot of the rate constant, k versus temperature for the TG pyrolysis of PP using FCC catalysts. The kinetic parameters were obtained by the Isoconversion method. 99 A plot of the rate constant, k versus temperature for PP pyrolysis using S-ECAT. The kinetic parameters were derived by the Isoconversion method 102 TGA simulation of PP measured at 10 C/min. Arrhenius parameters: E a = 250 kj/mol and ln A = 37 min -1. Isoconversion parameters: E a = 170 kj/mol and ln A = 27 min TGA simulation of PP sample containing 3 wt% of S-ECAT measured at 10 C/min. Arrhenius parameters: E a = 300 kj/mol and ln A = 46 min -1. Isoconversion parameters: E a = 203 kj/mol and ln A = 33 min TGA simulation of PP sample containing 8 wt% of S-ECAT measured at 10 C/min. Arrhenius parameters: E a = 172 kj/mol and ln A = 26 min -1. Isoconversion parameters: E a = 108 kj/mol and ln A =17 min TGA simulation of PP sample containing 13 wt% of S-ECAT at 10 C/min. Arrhenius parameters: E a = 142 kj/mol and ln A = 24 min -1. Isoconversion parameters: E a = 119 kj/mol and ln A = 20 min TGA simulation of PP sample containing 8 wt% of Fresh Fines measured at 10 C/min. Arrhenius parameters: E a = 53 kj/mol and ln A = 6.5 min -1. Isoconversion parameters: E a = 180 kj/mol and ln A = 34 min TGA simulation of PP sample containing 8 wt% of Fresh CAT measured at 10 C/min. Arrhenius parameters: E a = 120 kj/mol and ln A = 19 min -1. Isoconversion parameters: E a = 150 kj/mol and ln A = 25 min TGA simulation of PP sample containing 8 wt% of A-ECAT measured at 10 C/min. Arrhenius parameters: E a = 215 kj/mol and ln A = 33 min -1. Isoconversion parameters: E a = 154 kj/mol and ln A = 25 min xii

13 Figure 4.24 Arrhenius plot for the degradation of virgin PE 109 Figure 4.25 Figure 4.26 Figure 4.27 Figure 4.28 Arrhenius plot for the catalytic degradation of PE using 8 wt% Fresh Fines 109 A plot of the rate constant, k versus temperature using the kinetic parameters derived by the Isoconversion method in the TG pyrolysis of PE using FCC catalysts 111 Conversion of PP at 3 wt% of various FCC catalysts by dry Mixing 112 Conversion of PP at 8 wt% of various FCC catalysts by Dry Mixing 113 Figure 4.31 Repeated melt-mixing for 3 wt% and 8 wt% Fresh FCC Fines 114 Figure 4.32 Repeated Dry-Mixing for 3 wt% and 8 wt% Fresh FCC Fines 114 Figure 4.33 Comparison of the Melt mixing (MM) and Dry mixing (DM) TG analysis of PP degradation at 3 wt% of Fresh CAT 115 Figure 4.34 Figure 4.35 Figure 4.36 Comparison of the Melt mixing (MM) and Dry mixing (DM) TG analysis of PP degradation at 3 wt% of Fresh Fines 116 Comparison of the Melt mixing (MM) and Dry mixing (DM) TG analysis of PP degradation at 8 wt% of Fresh CAT 117 Comparison of the Melt mixing (MM) and Dry mixing (DM) TG analysis of PP degradation at 8 wt% of Fresh Fines 117 Figure 5.1 Schematic of twin-screw extruder 121 Figure 5.2 Figure 5.3 Schematic of the twin-screw reactor for the catalytic pyrolysis of polyolefin waste (melt zone not shown). 122 Cross section of a CRNI twin-screw extruder showing the apex width (w a ) and angle 126 Figure 5.4 DTA plot for the non-catalytic pyrolysis of PP 127 Figure 5.5 DTA plot for PP pyrolysis at 3 wt % S-ECAT 127 Figure 5.6 DTA plot for PP pyrolysis at 8 wt % S-ECAT 128 Figure 5.7 Extruder temperature profile at 500 RPM 144 xiii

14 Figure 5.8 PP conversion profile at 500 RPM 145 Figure 5.9 PP flow profile at 500 RPM 145 Figure 5.10 Extruder product flow profile at 500 RPM 146 Figure B.1 Arrhenius plot for the degradation of 8 wt % A-ECAT 159 Figure B.2 Arrhenius plot for the degradation of 8 wt % Fresh CAT 159 Figure B.3 Figure B.4 Figure B.5 Arrhenius plot for the catalytic degradation of PP using 3 wt% S-ECAT 160 Arrhenius plot for the catalytic degradation of PP using 8 wt% S-ECAT 160 Arrhenius plot for the catalytic degradation of PP using 13 wt% S-ECAT 161 Figure B.6 TG plot for the degradation of PP using 8 wt % A-ECAT 164 Figure B.7 TG plot for the degradation of PP using of 8 wt % Fresh CAT 164 Figure B.8 TG plot for the degradation of PP using 8 wt % Fresh Fines 165 Figure B.9 TG plot for the degradation of Virgin PP 165 Figure B.10 TG plot for the degradation of PP using 3 wt % S-ECAT 166 Figure B.11 TG plot for the degradation of PP using 13 wt % S-ECAT 166 Figure B.12 Isoconversion TG data analysis method. Plot of log β versus T -1 for degradation of PP using 8 wt % A-ECAT 167 Figure B.13 Isoconversion TG data analysis method. Plot of log β versus T -1 for degradation of PP using 8 wt % Fresh CAT 167 Figure B.14 Isoconversion TG data analysis method. Plot of log β versus T -1 for degradation of PP using 3 wt % S-ECAT 168 Figure B.15 Isoconversion TG data analysis method. Plot of log β versus T -1 for degradation of PP using 8 wt % S-ECAT 168 Figure B.16 Isoconversion TG data analysis method. Plot of log β versus T -1 for degradation of PP using 13 wt % S-ECAT 169 Figure B.17 Arrhenius plot for the catalytic degradation of PE using 8 wt% A-ECAT 169 xiv

15 Figure B.18 Figure B.19 Arrhenius plot for the catalytic degradation of PE using 8 wt% Fresh CAT 170 Arrhenius plot for the catalytic degradation of PE using 8 wt% S-ECAT 170 Figure C.1 DSC plot for the non-catalytic pyrolysis of PP 186 Figure C.2 DSC plot for PP pyrolysis at 3 wt % S-ECAT 186 Figure C.3 DSC plot for PP pyrolysis at 8 wt % S-ECAT 187 xv

16 SUMMARY The pyrolysis of Polypropylene (PP), Polyethylene (PE) and PP-Post consumer Carpet (PP- PCC) by various Fluid Catalytic Cracking (FCC) catalysts was studied by thermogravimetry (TG). All the FCC catalysts enhanced the pyrolysis of PP and PE. The reaction rates increased with catalyst fraction and a reduction in the catalyst particle size. On the other hand, the FCC catalysts performed poorly in the degradation of PP-PCC at 92 and 97 wt % of PP-PCC. This may be attributed to the waste composition of the PP-PCC. However, the pyrolysis of the waste PP was achieved at much lower temperatures than the virgin PP. In general, the fresh catalyst (fines) was found to be the most effective in both the polyolefin and PP-PCC pyrolysis. Furthermore, the rate law parameters, E a and A, for the pyrolysis reaction were estimated by conventional dynamic TG analysis. Despite the complex nature of polyolefin degradation, a single degradation step involving the solid polymer and the gaseous products was assumed. Although, there are many limitations and assumptions associated with conventional dynamic TG analysis, the derived rate equations for the different catalyzed reactions were used in developing a model for the catalytic pyrolysis of PP in an extruder. Based on the reactor model, basic processing costs were estimated for the catalytic pyrolysis of PP using various FCC catalysts at two screw speeds, 250RPM and 500 RPM. The lowest costs achieved were less than $1 per pound of product with the catalyst. xvi

17 CHAPTER 1 INTRODUCTION Due to population increase, the demand for plastic products has steadily increased over the last 40 years. Since plastics are non-biodegradable, they cannot be easily returned to the natural carbon cycle; hence the life cycle of plastic materials ends at waste disposal facilities. In 2003, the Environmental Protection Agency (EPA) reported that plastics constituted a large part of the municipal solid waste (MSW) generated in the US (26.7 million tons or 11.1% of the MSW stream) [1] Total MSW (Millions of Tons) Total Generated Total recovered Total Incinerated Total Landfill Landfill + Incineration Figure 1.1 Municipal Solid Waste Generation, Recycling, and Disposal in the United States [1] 1

18 The current and common MSW management methods used are incineration and landfilling. Figure 1.1 shows the trends for MSW generation and disposal over the last 40 years. Both methods pose negative environmental impact. Although, there are incineration methods in which energy is recovered, incinerators generally produce greenhouse gases which are postulated as sources of global warming [2, 3]. Similarly, landfilling poses the threat of methane emissions [4]. In light of these hazards, the EPA has improved federal regulations for landfilling by normalizing the use of liners in the landfill bed, ground water testing for waste leaks, and post landfill closure care; however, since waste plastics have a high volume to weight ratio, appropriate landfill space is becoming both scare and expensive [5]. Recycling and reuse of plastics has obvious benefits of decreasing the amount of waste plastics that end up in landfills; however, the overall recovery of plastics for recycling is relatively small. In 2003, only 1.4 million tons (3.9 percent of total plastics generated) of plastics were recovered for recycling [1]. The growing awareness in environmental concerns and the reducing landfill space have prompted research in alternative methods such as chemical recycling. Chemical or feedstock recycling involves processes that convert plastic waste into petroleum feedstock, preferably gasoline range fuel. Although a viable option, this method can be costly. Contributing to the high costs is the fact that waste plastics are indeed mixtures of different materials having different compositions and thus requiring different processing conditions. Moreover, the chemical recycling (or cracking) of plastics has to be combined with other technologies, such as MSW collection, categorization and pretreatment at the upstream end, as well as various separations and product recovery processes on the downstream end. Another prevalent issue with this method is the high energy input required as some post 2

19 consumer plastic waste require temperatures as high as 700 C [6]. By using effective catalysts, these temperatures can be significantly reduced, thereby reducing costs. Nonetheless, the catalyst cost can in turn affect the process economy considerably. From an economic perspective, reducing the cost even further will make feedstock recycling an even more attractive option. This option can be optimized by: Reuse of catalysts (this is common for reactor design in industry) The use of effective catalysts in lesser quantities. The first goal of this study is to investigate the catalytic decomposition of Polypropylene (PP), Polyethylene (PE) and post consumer carpet (PCC) using thermogravimetry (TG). FCC catalysts have shown great potential in the cracking of polyolefins; however, there is limited kinetic data on catalytic polyolefin pyrolysis using FCC catalysts in literature. Thus, kinetic parameters will be estimated by various methods described in literature for the TG decomposition of PP and PE. Furthermore, the catalytic effect of CaCO 3 in the decomposition of polypropylene-post consumer carpet (PP-PCC) will be studied by TG methods. CaCO 3 is used as filler in most tufted carpet and any catalytic potential of this carpet backing component may be useful in the recycling of PCC waste. Next, based on the outcome of the catalysts screening, kinetic data obtained by TG will be used in generating a reaction model for the cracking of PP in a twin screw, counter-rotating non-intermeshing (CRNI) extruder- reactor. Two key assumptions have been made in the process: (1) The extruder is modeled as a plug flow reactor and (2) the PP cracking products are assumed to be similar to those in literature. In addition to estimating PP conversion, the 3

20 model may be used as a cost estimation tool for the decomposition PP to gasoline range fuels using a twin screw CRNI extruder. An additional goal is to explore the effect of certain processing parameters in the TG degradation of polyolefins. First, the catalyst particle size in the TG degradation of PP and Polyethylene (PE) will be studied. Two particle sizes ( regular and fines ) of an FCC catalyst will be investigated in PP and PE pyrolysis. Furthermore, the mode of catalyst contact with the polymer will be investigated. Marcilla et. al. dealt with TG (catalytic pyrolysis) of polyolefin, where the polymer(s) and catalyst(s) are finely ground, mixed, and then directly analyzed with TG [3]. This method may not effectively disperse the catalyst(s) and polymer(s) uniformly. Inhomogeneous dispersion could lead to mass transfer limitations that are characteristic of heterogeneous catalysis. Therefore, to ensure that the catalyst(s) are uniformly dispersed within the polymer samples for effective degradation, an alternative mixing method is needed. This problem may be alleviated by mixing the catalyst in the polymer melt phase (melt mixing) prior to the decomposition reactions as is done in this work. 4

21 CHAPTER 2 LITERATURE REVIEW Pyrolysis is generally defined as the controlled burning or heating of a material in the absence of oxygen [7]. In plastics pyrolysis, the macromolecular structures of polymers are broken down into smaller molecules or oligomers and sometimes monomeric units. Further degradation of these subsequent molecules depends on a number of different conditions including (and not limited to) temperature, residence time, and the presence of catalysts as will be discussed in this review. 2.1 Thermal Pyrolysis of Polyolefins The non-catalytic or thermal pyrolysis of polyolefins is a high energy, endothermic process requiring temperatures of at least C [8-10]. In some studies, temperatures as high as C are essential in achieving decent product yields [6, 11, 12]. Thermal pyrolysis of both virgin and waste plastics as well as other hydro-carbonaceous sources has been studied extensively in the past. A good number of these thermal cracking studies are on PE [8, 13-27], polystyrene (PS) [10, 13-17], and PP [9, 14, 15, 18-32]. On the other hand, only a few have researched the thermal decomposition of other common plastics such as polyvinylchloride (PVC) ([33, 34]), polymethyl methacrylate [22], polyurethane [35], and polyethylene terephthalate [34]. 5

22 2.1.1 The HC Cracking Mechanism A thorough study on the mechanism for the thermal decomposition of polymers is presented by Cullis and Hirschler [36]. The four mechanisms proposed are: (1) End-chain scission or unzipping: Cracking is targeted at chain ends first, and then successively works down the polymeric length. Unzipping results in the production of the monomer. (2) Random-chain scission: Random fragmentization of polymer along polymer length. Results in both monomers and oligomers. (3) Chain-stripping: Side chain reactions involving substituents on the polymer chain. (4) Cross-linking: Two adjacent stripped polymer chains can form a bond resulting in a higher MW species. An example is char formation. The thermal pyrolysis of PP and PE is known to follow the random chain scission route, resulting in mainly oligomers and dimers [6]. This mechanism is illustrated for PE and PP in Figures 2.1 and 2.2, respectively. Peterson et. al. observed that PE decomposition by thermogravimetry yielded mainly 1-hexene and propene [37]. Figure 2.1 Random chain scission in polyethylene [37] 6

23 Similarly Peterson et. al. observed that in the thermal pyrolysis of PP, the main products were pentane, 2-methyl-1-pentene and 2,4 dimethyl1-heptene [37]. During degradation, methyl, primary and secondary alkyl radicals are formed, and by hydrogen abstractions and recombination of radical units, methane, olefins and monomers are produced [23]. Figure 2.2 Random chain scission in polypropylene [37] Whereas the pyrolysis of PP and PE is characterized by low monomer yields [2, 33, 38], PS is known to follow an end-chain mechanism or depolymerization steps resulting in mostly monomeric units as the main product, as illustrated in Figure

24 Figure 2.3 Unzipping mechanism in polystyrene [37] The polyolefin samples are typically degraded in a closed reactor/melting vessel and heated to a reaction temperature at which the polymer decomposes. A reaction time is allowed and over time, the degradation products (gaseous, liquids and solid) are collected and analyzed. Common methods for liquid products analyses include Infra Red (IR), Mass Spectroscopy (MS) and gas chromatography (GC). Whereas gaseous products are analyzed typically by Nuclear Magnetic Resonance (NMR), Fourier Transform Infrared Spectroscopy (FTIR), and GC/MS. Solid residues are identified by gel permeation chromatography. Several couplings of the aforementioned analytical methods are available, including FTIR/MS and GC/MS Thermal Pyrolysis Product Yields Liquid product yields greater than 82.5% and as high as 96% have been observed for PE thermal pyrolysis [26, 39]; however, these were obtained at high temperatures (greater than 420 C) and within a reaction time of approximately one hr. Invariably, the gaseous products obtained by thermal cracking are not suitable for use as fuel products, requiring further 8

25 refinin g to be upgraded to useable fuel products [40, 41]. Generally, thermal cracking results in liquids with low octane value and higher residue contents at moderate temperatures, thus making the process an inefficient process for producing gasoline range fuels [15, 29]. A few researchers have sought to improve thermal pyrolysis of waste polyolefins without employing the use of catalysts; however, these changes either yielded insignificant improvements or added another level of complexity and costs to the system [29, 42]. 2.2 Catalytic Cracking of Polyolefins Catalytic degradation of polymers has shown the greatest potential to be developed into a commercialized process. In comparison to the purely thermal pyrolysis, the addition of catalysts in polyolefin pyrolysis: Significantly lowers pyrolysis temperatures. A significant reduction in the degradation temperature and reaction time [43] under catalytic conditions results in an increase in the conversion rates for a wide range of polymers at much lower temperatures than with thermal pyrolysis [44-46]. This effect is illustrated in Figure 2.4. Narrows and provides better control over the hydrocarbon (HC) products distribution in LDPE [47, 48], HDPE, PP[49, 50] and PS [51, 52] pyrolysis. While thermal pyrolysis results in a broad range of HCs ranging from C 5 to C 28 [35], the selectivity of products in the gasoline range (C 5 -C 12 ) are much more enhanced by the presence of catalysts [32, 45, 53]. Increases the gaseous product yields. Under similar temperatures and reaction times, a much higher gaseous product yield is observed in the presence of a catalyst for PE [45, 54]. 9

26 Increases the product yield in the gasoline range whereas a purely thermal process will produce more light gas oils [55]. Zeolites in particular are known to enhance the formation of branched hydrocarbons and aromatics [56]. Oils obtained by catalytic pyrolysis contain less olefins and more aromatic content [43]. Table 2.1 summarizes the results obtained by Miskolczi et. al. for the batch pyrolysis of HDPE [57]. Results show that both yields and chemical compositions of the resulting products are changed by the catalytic effect. Obviously the extent of the catalytic effect will vary according to the catalysts inherent properties. Figure 2.4 Conversion obtained in the thermal and catalytic cracking of HDPE, LDPE and PP (400 C, 0.5 h, plastic/catalyst=50 w/w). Catalysts are Al-Beta and Al-Ti-Beta zeolites [53] 10

27 Table 2.1 Thermal versus catalytic pyrolysis of HDPE in a Batch reactor for 1 hr [57] Non- Catalytic Catalytic NCM a FCC b HZSM5 c Temperature, C Gases, wt% Liquid, wt% Coke/residue, wt% a A clinoptilolite b Equilibrium fluid catalytic cracking catalyst c Commercial H-formed ZSM-5 catalyst Thus, the dramatic effect of catalyzed decomposition of polymers has spurred a wave of research in the area of catalysis and polymer degradation. The effect of both novel and traditional catalysts has been extensively investigated. In many of these studies, solid acid catalysts have been widely used and have shown much greater cracking activity over nonacidic catalysts [55]. Solid acids are particularly important in the petroleum industry where many reactions proceed via acid-catalysis e.g. paraffin isomerization, catalytic cracking, reforming and alkylation Catalytic cracking pathway Reaction products are largely determined by carbenium ion chemistry (isomerisation, chain/beta-scission, H-transfer, oligomerisation/alkylation) which is influenced by acid-site strength, density and distribution [58]. The acid strength of solid acids is characterized by both Brønsted and Lewis acid sites; however, the presence of Brønsted acid sites have been observed to favor the cracking of olefinic compounds [43]. A study of the Brønsted and Lewis acid sites in polyolefin cracking has been reviewed by several authors [59-62]. Furthermore, in the case of crystalline solid acids, the majority of the acid sites are believed 11

28 to be located within the pores of the material, such as with zeolites [63]. Thus micro-porosity of porous solid acids is an important feature in assessing the level of polyolefin cracking over such catalysts. Molecular sieves, such as silica alumina [15, 55, 64, 65], zeolites [66-69], and MCM-41 [70-72], are among the most commonly researched solid acids in plastic waste pyrolysis. Other catalytic materials such as clays have been scarcely investigated [73]. Generally, the level of the catalyst activity in polyolefin pyrolysis increases with increasing number of acid sites [43]. Thus it is common knowledge that zeolitic catalysts achieve higher conversion than non-zeolitic acid catalysts [38, 55, 56] Zeolites in polyolefin pyrolysis Zeolites are usually described as crystalline aluminosilicate sieves having open pores and ion exchange capabilities [58]. This characteristic makes them useful in common applications, such as water purification, and in catalysis of a wide spectrum of organic reactions with applicability in the petroleum industry [74]. More specifically, a zeolite consists of a rigid tetrahedral network of SiO 4, AlO - 4 or PO + 4 frameworks [58]. Combinations of the latter two frameworks are found in most common zeolites, resulting in structures that stack in three dimensional shapes forming cavities of varying size and shape as shown in Figure 2.5. The resulting crystal structure and properties depend on the direct synthesis conditions, such as temperature, gel composition, templating agent and the presence of alkaline cations [75]. Cundy and Cox present a thorough review on the hydrothermal synthesis of zeolites [76]. Zeolites were introduced into the refinery about 40 years ago. These molecular sieves combine high acidity with shape selectivity, high surface area and high thermal stability to 12

29 catalyze a variety of hydrocarbon reactions including polyolefin cracking [77]. The reactivity and selectivity of zeolites as catalysts are determined by their high number density of active sites which are brought about by a charge imbalance between the silicon and aluminum atoms in the crystals, causing the zeolite to have an overall charge balance of negative one [58]. MFI (ZSM-5) FAU (Y-zeolite) LTA (Zeolite A) Figure 2.5 Structures of some common zeolites used in polyolefin pyrolysis [78]. For this reason, they can catalyze many hydrogen transfer reactions. Moreover, because of their high thermal stability, zeolites can be regenerated by burning off the polymer on the surface and in the pores at very high temperatures. Thus, they are currently used as catalysts in the fluid catalytic cracking (FCC) processes of the petroleum industry [79]. A number of reviews on the different uses of zeolite catalysts in the petroleum refining industry are available in texts [80-84]. Aside from high acidity, the chemistry of zeolite catalyzed reactions is significantly influenced by their pore size, which is typically in the range of 4-13 angstroms. Small pore zeolite such as Zeolite A (~ 4 Å) have an 8 member oxygen ring and would typically allow small molecules such as olefins and alcohols to pass through. On the other hand, larger pore 13

30 zeolites such as zeolite-y (~ 7.4 Å) and mordenite (~6.7 Å) will tend to allow larger molecules to pass through their pores (See Figure 2.5). Therefore, reaction products are influenced by the shape selectivity of the zeolite catalyst [77, 85]. The shape selective mechanisms for diffusion in porous materials are known to fall under four major groups: Reactant state selectivity: Only sterically unhindered reactant molecules (reactants with smaller kinetic diameters) are permitted through the catalyst pores for further reaction. Product selectivity: Within the intrazeolitic pore system, the exit of certain products is sterically preferred over the other possibilities. A key example is the paraselectivity of HZSM-5 for the alkylation of toluene with ethylene [85]. Transition state selectivity: The intrazeolitic pore system provides different steric hindrances for the various transition states. It is expected that the transition state with the least resistance will in effect go on to form the main product. Molecular traffic control: The pores are important in determining the kinetics of the reactions in zeolites; however, the diffusion of either products or reactants may be improved or hindered by the global zeolite cage system such as the dimensionality of the pore system (1-D, 2-D, 3-D). The degradation of large olefinic molecules occurs over the surface of these catalysts, forming smaller molecules that can be permitted into the pores of the zeolites for further cracking and selectivity [86, 87]. Diffusion of these cracked molecules within the zeolite is 14

31 greatly influenced by pore size constraints and depends on the pore and channel configurations (since the size of the inner channels and HC molecules are about the same). This type of diffusion is commonly referred to as configurational diffusion. Diffusion in the Knudsen regime may also occur in some zeolites; however, it is believed that configurational diffusion dominates [88, 89]. Due to their effectiveness in the cracking of crude oil and petroleum derivatives, they have also been extensively cited in literature as effective catalysts in the study of catalytic polyolefin pyrolysis. Amongst the numerous kinds of zeolites investigated in polyolefin pyrolysis, (Beta [90], USY [91], ZSM-11 [72], REY [40, 92], Mordenite [69, 93]), ZSM-5 is the most commonly used. Conversely, with polystyrene, solid bases are observed to be the most effective [94] ZSM-5 in Polyolefin Pyrolysis In literature, ZSM-5 has performed better than most other zeolitic compounds and parent silica alumina catalysts in polyolefin pyrolysis studies [95, 96]. The high catalytic activity of ZSM-5 is attributed to its strong Brønsted acid sites which is closely related to the presence of Al in the framework (a higher concentration of Al denotes increased number of acid sites); hence, the activity of a zeolite sample may be enhanced by varying its gel composition [45]. Other properties of zeolites can also be enhanced by ion-exchange, as well as by molecule impregnation of reactive species into the zeolite framework [72, 97]. It is widely observed that zeolites favor the production of gaseous products and aromatic compounds in fuel feedstock recycling (See Tables A-1, A-2 and A-3 in Appendix A). The high yield of gaseous products could be attributed to the over-cracking nature of these 15

32 microporous solid acids [73]. Microporous zeolites such as ZSM-5 exhibit high gas yields with higher aromatics and napthenes whereas mesoporous solids such as silica alumina and MCM-41 show high liquid yields with high olefinic content [38]. This is because the diffusion of molecules larger than monomethyl aliphatics into micrporous ZSM-5 is restricted and the reaction of molecules with critical pore diameters greater than 6Å is severely diffusion limited [98]. On the other hand, because of the relatively small pore size of ZSM-5, the interaction between the catalytic surface of ZSM-5 and the reactants is larger resulting in a higher conversion of linear olefins and a higher production of low MW compounds such as ethane. In one study, medium pore zeolites, ZSM-5 and Mordenite, formed significantly more olefins whereas pyrolysis over large-pore zeolites, Y and Beta zeolites, yielded mainly alkanes with less alkene and aromatic content [68]. In addition, significantly more lighter hydrocarbons (C 3 -C 6 ) were formed with ZSM-5 than with largerpore zeolites such as zeolites-y [68, 99]. Similar observations were made by Bagri et. al. who compared ZSM-5 and Y-zeolite in polystyrene pyrolysis [51]. They found that the former yielded higher gaseous products whereas Y- zeolites resulted in products having a higher aromatic content. A monomolecular and bimolecular cracking mechanism have been proposed to explain the wider product distribution by USY over ZSM-5 respectively [85, 100, 101]. The monomolecular mechanism is based on the assumption that the micropores of ZSM-5 are permissive to only mono-methyl compounds; hence, higher selectivity towards C 1 -C 3 species. In the pyrolysis study of HDPE, Manos et. al. observed that the use of US-Y zeolites gave products in the C 3 and C 15 range [102]. It was also observed that isobutane and isopentane were the main gaseous products whereas the liquid fraction was rich in alkanes [68, 102]. 16

33 Cardona et. al. studied the catalytic pyrolysis of polypropylene over various US-Y catalysts having different pore sizes [103]. In this study, the selectivity to gases decreased with increasing pore size, unlike the total acidity of the catalysts that seemed unrelated to cracking activity [103]. As such, the pore size of the catalyst is key in controlling the diffusion of the cracked intermediates for the catalytic degradation of waste plastics. Based on these observations, a catalyst with bi-modal pore size distribution such as those found in fluid catalytic cracking (FCC) catalysts may be used to optimize the liquid product distribution [104] FCC catalysts in Polyolefin Pyrolysis FCC catalysts have been employed on an industrial scale in the petroleum refining industry and were developed mainly for cracking heavy oil fractions from crude petroleum into lighter and more desirable gasoline and liquid petroleum gas (LPG) fractions [79]. The feedstock products fall under four major classes of HCs: Paraffins, Olefins, Naphthalenes and Aromatics (PONA distribution). Gasoline range fuels consist of paraffin and olefins in the C 5 -C 12 range [41]. Within aromatics, products of polyolefins, especially polystyrene, are grouped as BTX (benzene, toluene, and xylene). Recent reviews on the FCC process can be found in literature [79, 105, 106] The composition of an FCC catalyst An FCC catalyst is composed of zeolitic crystals and a non-zeolitic acid matrix (commonly silica alumina and a binder) [74]. Zeolite-Y is still the primary component of FCC catalysts for over 40 years because of its high thermal stability and product selectivity [77]. The alumina matrix, clay and binder serve to provide both mechanical and thermal stability 17

34 needed for HC cracking in FCC - regeneration unit cycles [74]. FCC catalysts are commercially available in two forms: powder or pellets. FCC powders typically range in the lower microns whereas the fluidizable pellets are typically 60 µm diameter. Various preparation methods for FCCs as well as the interaction between matrix and zeolite on catalytic activity have been investigated [107, 108] Performance of FCC catalysts in polyolefin pyrolysis Compared to the other common solid acids, FCC catalysts have only been recently studied in the cracking of commodity plastics. Marcilla et. al. conducted a thermogravimetric study of PP and PE mixtures in which the FCC catalyst was the most effective over ZSM-5 and Y- zeolite [109]. Similarly, Miscolkzi et. al. observed that the yield of gaseous products increased in the order: thermal cracking < clinoptilolite < FCC < HZSM-5 catalyzed cracking, while the yields of liquid products increased in the order of thermal cracking < clinoptilolite < HZSM-5 < FCC catalyzed cracking for the cracking of HDPE [57]. It is generally observed that FCCs result in more gasoline range products than did ZSM-5 and + Y-zeolite [57, 110]. In addition, the conversion of C 7 n-olefins and the production of ethene is higher in ZSM-5 than with FCCs [51, 100, 110]. This could be attributed to the smaller and thus restrictive micropores found in ZSM-5. On the other hand, the bimodal pore size distribution, and mild acidic properties of FCCs allow for the formation of more parrafins [104]. 18

35 Equilibrium FCC Catalyst (E-CAT) Equilibrium catalysts are used FCC catalysts with different minute levels of metal contamination but still have value. These metals are typically vanadium, nickel, iron, sodium. Also entrapped in the catalyst pores, is coke (carbon). Before reuse, the spent catalyst typically undergoes a number of burn-offs in regeneration units at temperatures as high as 700 C [103]. This is done to remove some of the entrapped carbon, therefore diminishing ECAT s catalytic activity [111]. Despite their diminished activity, these catalysts are still significantly effective in polyolefin pyrolysis compared to other acid catalysts. De la Puente et. al. studied the catalytic pyrolysis of polystyrene over ZSM-5, Mordenite, sulfur promoted zirconia and an equilibrium FCC catalyst [112]. In their study, the spent FCC catalyst resulted in higher yields of desirable products such as ethylbenzene [112]. Lee et. al. observed that at 430 C and in a semi batch reactor, the pyrolysis of waste HDPE using an ECAT showed high cracking activity, yielding better liquid gasoline range products than in purely thermal cracking [113]. Cardona et. al. compared the gasoline range selectivity of various silica alumina and Y-zeolites with spent FCC over PP and found a comparable gasoline range yield (greater than 70%) [103]. Table 2.2 summarizes the results published by Ali et. al., comparing the product selectivity of fresh and equilibrium catalysts [101] Coke Formation in FCC Catalysts Coke formation is a common problem with FCC catalysts in industry and in polyolefin pyrolysis research. Coke consists mainly of heavy aromatics formed during the polyolefin cracking process. 19

36 Table 2.2 Product distribution for degradation of HDPE in a fluidized bed reactor using fresh and equilibrium FCC catalyst at 360 C (C/P loading = 2:1, reaction time = 30 min) Yield (wt% of feed) Fresh FCC E-Cat a Gas Liquid Coke Residue Gas & Gasoline breakdown C1-C C5-C BTX b a E-cat has < 50 % BET surface area of fresh sample. ** BTX: Benzene, toluene and xylene content Due to the strong binding nature of these poly-aromatics, the activity of a catalyst drops with increasing coke content [114, 115]. It is also believed that even at low coke deposition, the strongest acid sites are involved in the coke formation [116, 117]. Catalytic coke is formed when side reactions occur in the larger pores of FCCs forming these high MW poly-aromatics. These molecules tend to be too large to escape through the pore opening of the FCC catalyst resulting in a deactivation of the catalyst over time. Unlike FCCs, microporous catalysts such as ZSM-5 show low coke deposition due to its small pore system which does not accommodate the formation of large molecules such as coke [38, 118, 119]. De la Puente et. al. observed that a FCC catalyst yielded more coke than ZSM-5 [112]. 20

37 On the other hand, coking may be as a result of poisoning of the catalysts by the metals found in the FCC unit such as nickel, iron and vanadium [120]. Bayraktar et. al. used analytical methods such as AFM, SEM-EDS, XPS and optical microscopy to show the role of Fe in FCC coking and catalyst deactivation [121]. Figure 2-6 Equilibrium FCC sample containing high levels of coke (Ni: 2600; Va: 6700; Fe:7000 ppm. A substantial amount of Fe deposits were found within the white rectangle. Magnification: 750X [122] Conversely, Guisnet et.al. have studied that the entrapped coke, which are themselves active species, can participate in certain catalytic reactions [123] Effect of Polymer Type on Product distribution PONA distributions of FCC catalyzed decompositions show that the olefin yield far exceeds the yield of paraffins, naphthenes, or aromatics (PNAs) in the pyrolysis of PP and HDPE [101, 124, 125]. Lee et. al. also showed that the catalytic degradation of waste LDPE produced more paraffins and aromatics than those of waste HDPE and PP [125]. Marcilla et. al. investigated the pyrolysis of different PE grades (LLDPE, HDPE, LDPE) by thermogravimetry. They observed slight differences in their decomposition behaviors but only in the presence of the catalyst (MCM-41) [126]. Conversely, PS pyrolysis exhibits high yields of aromatics, as high as 97 wt% of liquid product, far exceeding those obtained with 21

38 PE or PP (< 20 wt % of liquid yield) [22, 112, 125, 127]. Consequently, very low yields in PNAs are observed. This is attributed to the polycyclic nature of PS and the thermodynamic challenge posed in converting cyclic compounds to aliphatic chains or alkene compounds [112]. A closer look at the aromatic yield in many of these catalyzed reactions reveals that, the product selectivity is higher for benzene, toluene and ethyl benzene unlike in thermal pyrolysis, where the main product is styrene [6, 22, 112, 127, 128]. This clearly indicates the similarity and variance in the cracking mechanisms among these three polyolefins Effect of Particle/Crystallite Size on Product Distribution The effect of catalyst particle size has only been sparsely studied in literature. You et. al. investigated the effect of particle size of MFI zeolites on the catalytic degradation of polyethylene wax and found that whereas conversion decreased with particle size, product quality increased [87]. Furthermore, particle sizes in the nano-range have been investigated. Serrano et. al. reported conversions as high as 90%, temperatures less than 350 C for the cracking of PP, LDPE and HDPE using nano-crystalline ZSM-5 [129]. Aguado et. al. observed similar results in the batch pyrolysis of PP and LDPE mixtures using nano-hzsm5 [99]. Based on these results, it can also be deduced that nano-zsm-5 catalyzed reactions result in very high gas yields in the range of C 3 C 6 products, and apparently in much higher concentrations than is observed with micron-sized ZSM-5. These nano-sized particles are this effective because of their increased surface area. Conversely, high surface area combined with a very small pore system poses great difficulty in achieving decent amounts of gasoline range products in the C 5 -C 12 range. Moreover, the nano-catalyst selectivity to liquid products is also very limited [99, 129]. This could be resolved by investigating the particle size effect with catalysts that are selective to gasoline range liquid products such as FCC catalysts. 22

39 Costa et. al. found that a submicron base Y-zeolite for their FCC catalyst formulation showed a reduction in cracking of gas oil but showed a low selectivity for coke [108]. On the other hand, Tonetto et. al. observed that the effect of zeolite crystallite size on conversion and product distribution depended on the size of the decomposed hydrocarbon molecules [130]. The processes used in both studies, including the synthesis and embedment of sub-micron into an FCC catalyst seem both labor intensive and costly procedures. Subsequently, one may infer that an easier and economical approach might be to consider varying already formulated FCC catalysts with particle sizes ranging in the sub-microns. The effect of FCC catalyst fines on PP and HDPE pyrolysis will be discussed in this thesis Process Design Thus far, the effects of catalyst and polymer type on the resulting product distribution in polyolefin pyrolysis have been discussed. Literature shows that the distribution can also be affected by other process parameters such as the means of polymer and catalyst contact during degradation, reactor type, feed composition (virgin/waste plastic) and degradation process conditions. To avoid a lengthy bibliography, only the most recent (after 1999) and relevant works will be discussed in this review Catalyst Contact Mode One may be able to investigate the catalytic steps involved in polymer degradation by considering different modes of catalyst introduction to the polymer feed. Sakata et. al. investigated two modes of contact in the batch pyrolysis of PP using various solid acids: liquid phase contact and vapor phase contact [55]. For the catalytic degradation in the liquid phase contact, both catalyst and polymer are placed in the reactor and heated to the operating temperature. Whereas, with the vapor phase contact mode, the polymer is first 23

40 thermally degraded into HC vapors and then contacted with the catalyst. It was observed the HC vapors underwent further cracking in the vapor phase whereas the product yield in the liquid or melt phase contact did not differ significantly from that obtained by purely thermal degradation of PP [55]. In this study, two contact modes will also be investigated in the TG pyrolysis of PP using various FCC catalysts: melt mix and dry mix Reactor Type A wide range of reactors have been used on a lab-scale in polyolefin pyrolysis. The reactor set-ups investigated thus far fall under one of the following categories: Batch, Continuous flow (CFR), modifications or combinations of either of the aforementioned Batch and Semi-Batch Reactors [43, 45] A common variable in batch and semi-batch operations is nitrogen which is used for the continuous removal of volatiles from the reactor vessel. The products are then collected by passing the vapors through a condensation system. Most are made out of pyrex or stainless steel. Some of these works are tabulated in Table A-1 of Appendix A. A key disadvantage with this is the high reaction times observed. Furthermore, under batch operation, it seems that the potential of a catalyst is minimized with similar product yields to thermal at similar conditions. From an industrial viewpoint, continuous reaction systems are preferred to batch set-ups for operational reasons Fixed Bed Semi-Batch reactor [45, 131] Polymer and catalysts samples are heated separately and reacted by vapor phase contact. Degraded polymer fragments are carried to the catalyst bed/mesh by a carrier gas, in most cases N 2. Typically the catalyst bed is heated to a higher temperature than the polymer bed. 24

41 Fluidized bed batch reactors [ ] Riser simulator reactors are fluidized batch reactors, specifically designed to simulate similar conditions found in a catalytic riser reactor used in the FCC process. It is adapted for liquid phase catalytic reaction, in which heat from the catalysts could vaporize the melt polymer feed while simultaneously cracking the resulting hydrocarbons Continuous Flow Reactors (CFRs) [96, 135] More recently, researchers have moved focus towards reactors with greater feasibility in the industrial arena such as fluidized bed reactors which mimic the FCC unit in the petroleum industry. Generally, CFRs are characterized by much shorter residence time (less than a few seconds to a few minutes), improved uniformity and dispersion. Most of the more recent works in polyolefin pyrolysis are on fluidized bed reactors (See Table A-2). The use of continuous flow reactors in polyolefin pyrolysis prior to 1998 has been discussed [135]. The University of Hamburg, in particular, has done a lot of research in feedstock recycling from waste plastics using FCCs, and has subsequently developed the Hamburg process which makes use of an indirectly heated fluidized bed [22, 127]. During catalytic cracking, quartz sand is replaced by the respective FCC catalyst as packing material. Amongst the various catalysts investigated, FCCs produced the most decent liquid yields in PE pyrolysis as shown in Table A.2. Unlike a batch reactor, a fluidized bed reactor is suited for pyrolysis because it provides very good heat and material transfer rates hence generating largely uniform products. However, the disadvantages are many and include: 25

42 Broad residence time distribution of solids due to intense mixing. Attrition of bed internals and catalyst particles. Difficulty in scale-up. Defluidization problems [136]. Requires large amounts of catalysts. Low liquid yields due to over cracking (See to Table A.2) On the other hand, other continuous systems, such as the three-step continuous flow pyrolysis process involving a pre-heat, cracking reactor and separation zones, have been investigated by a few [26, 35, 42, 47]. In this method the polymer is first pre-heated to a molten state in a CFR such as an extruder and driven into the reactor where it is further cracked at elevated temperatures. Table A.3 summarizes recent and relevant polyolefin pyrolysis works employing CFRs Effect of Feed Composition Many have demonstrated that plastics waste can indeed be converted to useful chemical feedstock by both non-catalytic [6, 33, 44, 124, 125, 133, 137, 138] and catalytic pyrolysis [35, 38, 101, 103, 133, 139, 140]. The present issues are the necessary scale up, minimization of waste handing costs and optimization of gasoline range products for a wide range of plastic mixtures or waste. In addition, controlling the product distribution is still an issue with waste and mixtures. Waste contents like PVC [9, ] and biomass [ ] do have an influence on the pyrolysis products. 26

43 In general, the decomposition of polyolefin mixtures occurs roughly in the same range as their virgin counterparts ( C). However, waste polyolefins may degrade at slightly lower temperatures and achieve higher conversions than the respective virgin polyolefins [44, 113, 139, ]. As with virgin plastics, the addition of catalysts in waste pyrolysis greatly influence product yields and conversion rates; however, the disparities between waste and virgin polyolefin pyrolysis lie mainly in the resulting product compositions [35, 38, 52, 109, 124]. It is clear that during pyrolysis, interactions between the different materials in a waste feed have a significant effect on the selectivity of specific liquid and gaseous product components as shown in Table 2.3 [150]. Typically, PE pyrolysis favors mostly the formation of paraffins; however, upon increasing its PS or PP content, the yield of aromatic and alkenic products is greatly enhanced, thus improving its octane value [124, 140]. Due to the radicals formed during PS decomposition, the conversions of PP and PE are improved by PS addition [52, 149, 151]. Conversely, PS decomposition seems to be immune to effects by either of the other polyolefins. Table 2.3 Waste versus virgin pyrolysis of HDPE using ZSM-5 and under similar operating conditions. (All experimental parameters are very similar between both references) Yield (wt% of feed) Virgin HDPE ([101]) Waste HDPE ([38]) * Gas Liquid Coke Residue Gaseous product breakdown C1-C C5-C BTX Approximate waste composition: 38 wt% HDPE, 24wt% LDPE, 30wt% PP, 7wt% PS, 1wt% PVC 27

44 Marcilla et. al. found that an FCC catalyst performed better than ZSM-5 and USY in the catalytic pyrolysis of PP-PE mixtures [109]. On the other hand, Lin et. al. found that larger pore zeolites (MOR and USY) and non-zeolites (SA, MCM-41) yielded much more coke and residue in comparison to ZSM-5 [38] Effect of other Process Parameters The effect of other process parameters such as reaction temperature, pressure, reaction time and catalyst loading has been investigated in literature. These are summarized in Table

45 Table 2-4 Influence of certain process conditions in polyolefin pyrolysis Process Parameter Temperature [6, 122, ]. Results Conversion increases with temperature resulting in decrease of aliphatic content. Dermibas et. al. observed that gaseous products (C2-C4) and liquid products (C5-C9) increased and decreased with temperature respectively [6]. Effect of the catalysts on the yields and structure of products becomes less significant with increasing temperature [57, 95]. Pressure Murata et. al. demonstrates the inverse relation of pressure to temperature in the pyrolysis of polyethylene [155]. Residence time Key parameter in fluidized bed reactors. Generally conversion increases with residence time. [11, 65, 153, 154, 156] Miskolczi et. al observed that the catalyst activity of HZSM-5 and an FCC catalyst decreased with increasing cracking time in the pyrolysis of HDPE waste. Effect of residence time on product yield is more pronounced at lower than higher temperatures Catalyst loading Conversion increases with catalyst loading. [99, 154, 157] 29

46 2.3 Kinetic Studies in Polyolefin Pyrolysis Kinetic models are important in the reactor design and scale up of industrial processes. Within the extensive work done in polyolefin pyrolysis, the development of kinetic models describing the cracking process has been narrowly researched. Nevertheless, a wide range of models have been developed. Only the relevant and recent works have been cited in order to portray the current status of this field. The kinetic models in literature are broadly based on either representing the thermogravimetric method or detailed reaction mechanisms Kinetic Models Based on Thermogravimetry (TG) In addition to the assessment of polymer stability and compositional analysis, thermogravimetry is used to estimate the degradation kinetics of polymers. These kinetic models are based on the analysis of the weight loss curves obtained during thermal gravimetric analysis of the polymeric samples. The analysis is based on the fact that polymer decompositions are typically heterogeneous (solid state) reactions, and so the rate of conversion or weight loss (X) is a linear function of a temperature dependent rate constant, k, and the weight loss function, f (X), is described as follows: dx dt E = kf ( X ) = Aexp a f ( X ) (1) RT Ea ln k = ln A (2) RT 30

47 Wo W X = ; [ X ] f o (3) Wo W f where A is the pre-exponential factor, Ea is the activation energy, W is the weight of the sample (polymer and catalyst) at time t, and the subscripts o and f represent times at the beginning and end of the degradation respectively; f (X) is a weight loss function (WLF) which describes the dependence of the rate of conversion to the amount of residual polymer. For a simple thermal degradation process of polymers, most describe the WLF as the power law function shown in Equation 4. n is the order of the degradation reaction Thus, Equation 1 becomes f ( X ) ( X ) n = 1 (4) dx dt E = A exp a ( 1 X ) n (5) RT The determination of degradation kinetics by TG can be approached in several ways: Isothermal TG measurement As the name suggests, the temperature is set to a constant during the entire decomposition process. For several measurements, the maximum rate of conversion can be obtained for each temperature and the logarithmic form of Equation 5 can be used to obtain the kinetic parameters (E a, A and n) as shown in Equation 6 [158]. 31

48 dx Ea ln = ln A + n ln ( 1 X max ) (6) dt RT max On the other hand, the kinetic parameters can also be calculated from an integrated form of Equation 5 [159]. ( a) g Ea ln () t = ln + A (7) RT where g( a) n 1 X dx (1 X ) 1 = = { ln(1 X ) for n = 1 }; for n 1 0 n ( 1 X ) n 1 The main problem with this method is that it is time consuming and cumbersome requiring extrapolations for temperatures above the equipment range [160] Dynamic TG measurement Dynamic (non-isothermal) measurements provide cumulative weight loss data at each linear rising temperature controlled by the chosen heating rate, β ( /min). Many methods that use this concept have been proposed. The most commonly used are the model-free isoconversion methods by Ozawa [161], Flynn and Wall [162], Kissinger [163] and Friedman [158]. The first three are very similar in that they can be used to estimate the kinetic parameters directly from weight loss and temperature data (integral isoconversion methods) whilst the Friedman method is of a differential form (derivative isoconversion method). o 32

49 Using a change of variables manipulation, Equation 5 c an be expressed in terms of β. dx A E dt ( ) a n = exp (8) d 1 X β RT Therefore, E a and A can be estimated from TG data collected at different heating rates. The Ozawa-Flynn-Wall (OFW) method is derived from the integration of Equation 8 and has been incorporated into the American Standard for Testing Materials [164] as well as several commercial software packages [165] on most TG instruments. A detailed critical analysis of all the aforementioned isoconversional methods is discussed in literature [ ]. Drawbacks with this method lie in the assumptions made. First, it assumes a first order reaction model for thermal pyrolysis of plastics, even though the reaction order changes with conversion [136]. Second, because of an approximation that was made by Doyle in the integration of Equation 8, only conversion values in the low range can be used in the analysis [169] DTG Peak property method (DTG-PPM) Whereas, OFW is an integral method, Equation 8 may also be used in its differential form, also called a derivative thermogravimetry (DTG) curve. Useful kinetic data can also be obtained indirectly from TG data such as in DTG (derivative thermogravimetry) curves expressed by Equation 8. The DTG peak properties (peak temperature, peak conversion and peak height) are known to have a close relationship with global kinetic parameters, E a and A [163, ]. Park et. al. gives a detailed analysis on the role of peak parameters (peak 33

50 temperature, height) in determining kinetic parameters for thermal reactions [172]. In a recent study, Kim et. al. estimated n from peak properties of a DTG curve in a HDPE pyrolysis study and showed close similarities to results obtained by other conventional methods [175]. This method is prone to experimental errors during estimation of the exact peak properties, as well as in the choice of a weight loss function, although, (1-α) n is commonly used [172]. Unlike the model free methods described above, there exist model-fitting methods which require knowledge of the reaction mechanism or input parameters from other forms of analysis, including model free methods. These include the Freeman-Carol [176], Sharp- Wentworth [177], and Coats- Redfern [178] methods Modulated Thermogravimetry (MTG) This is a relatively new and model-free approach for obtaining TG kinetics. The temperature profile in MTG uses a linear temperature ramp controlled by an oscillatory temperature program that induces an oscillatory mass flow which is proportional to the physical properties of the polymer specimen [160]. The activation energy of a decomposition reaction may be calculated using Equation 9. E a = (R[T 2 av (0.5T amp ] 2 L)/T amp (9) Mamleev et. al. recently presented a critical comparison of the MTG with other model-free methods (OFW, Friedman) as well as a model-fitting method based on MTG [167] Factors affecting kinetic parameters estimated by TG 34

51 While certain kinetics factors, such as the surface area of the reacting sample, do not seem to affect the value of Ea, others, such as the extent of degradation, significantly influences the approximation of E a [37]. Over a conversion range of 0-95%, Peterson et al. found that E increased by approximately 60 % for PE, 67% for PP and negligibly for PS [37]. A summary of kinetic parameters obtained in more recent literature is displayed in Table 2.5 (Kinetic studies for the non-catalytic pyrolysis of polyolefins in literature prior 1999 has been reviewed [37]). a 35

52 Table 2.5 Literature values of the activation energy (E a ), reaction order (n) and preexponential (A) for degradation of PE, PP, and PS. Polymer-catalyst E a (kj mol -1 ) A n Method Refs. HDPE- none * DTG- PPM [175] none * Isothermal [179] none * OFW [179] none nd 1 OFW [180] almcm nd 1 OFW [180] none 221 nd Reactor kinetics [26] LDPE- almcm nd nd nd [181] PP- none * Coats-Redfern [84] none a 0.01 Friedman [15, 182] none 216 nd nd Reactor kinetics [26] PS- none a 0.32 Friedman [15] none * CSPR [136] none * FBR [13] none 208 Reactor kinetics [26] nd not determined; a may have reported lna although papers report A 36

53 Table 2.5 shows a wide variation between the activation energies and pre-exponential factors found by the various authors. These differences may be due to the different properties and characteristics of the polyolefins used in the various works as well as differences in process conditions from which kinetic data were estimated. Furthermore, a very wide range of kinetic models have been developed using these methods. Publications before 1999 for the study PE pyrolysis by both isothermal and non-isothermal methods have been documented and their applicability discussed by Ceamanos et. al. [179]. More recently, only a few researchers have used one or two of these TG kinetic models in modeling polyolefin pyrolysis [84, 146], comparing results obtained by isothermal and nonisothermal TGA methods. Discrepancies were found amongst the studies due to different reaction definitions [179, 182, 183]. In a fairly recent paper, Marcilla et. al. model the thermal and catalytic steps in PE pyrolysis by TG. They included a Michaelis-Menton type constant in their kinetic reactions to describe the saturation effect on catalyst activity commonly observed in polyolefin pyrolysis; however, results from a latter paper deemed this parameter unnecessary [71]. TG is a quick and easy way to obtain the global kinetics of a degradation reaction, however, a critical shortcoming is the heat transfer limitations within the sample, as there is a noticeable influence of heating rates on the activation energy [136, 175]. This problem is typically minimized by using considerably thin and small amounts of sample [164]. 37

54 2.3.2 Kinetic Models Based on Mechanistic Modeling Kinetic Models based on Reaction Mechanisms and Elementary Reactions With these models, the reaction rates are expressed in terms of the concentrations of the species involved in the elementary steps [184, 185]. Johannes et. al. describe thermal cracking using steps of individual reactions on the basis of well-known kinetic equations and experimental data obtained for the decomposition of LDPE [186]. These kinetic models typically involve a great number of kinetic parameters and does not account for secondary reactions that take place during polyolefin pyrolysis Kinetic Models based on Multi-step Reactions (Lumping System) These kinetic models involve multi-step reactions accounting for secondary reactions that take place during pyrolysis. Only a few of these mechanistic models are available in literature, especially for catalytic cracking. They are typically complex, consisting of a large numbers of kinetic parameters and reactions that describe and seek to quantify the polyolefin decomposition products [10, 28, 151, ]. These models are based on predictions of a typical free radical chain mechanism including reaction steps such as chain-end and random scissions, inter- and intra-molecular hydrogen abstractions, unzipping, backbiting, radical recombination and disproportionation reactions [85]. They are developed based on population balance equations solved by moment techniques described rigorously in literature [190, 191]. This theory is used to describe molecular weight distributions of species (macromolecules and radicals) in a polymer mixture, also called distribution kinetics [ ]. In addition, most of these models are based on a well known lumping system in which individual components of the pyrolysis product stream are grouped into measurable compound classes such as long-chain-olefins, olefins, parrafins, BTX, carbenium ions, coke 38

55 and much more [86, 92]. Reactions between these kinetic lumps are also modeled based on experimental data [184]. In a recent study, Lin et. al. improved the conventional lumping scheme to account for the catalytic degradation of HDPE and PP over fluidized acidic catalysts in a fluidized bed reactor [86]. In a latter study, Cardona et. al. presented a kinetic model for the pyrolysis of PP in a stirred batch reactor and accounted for thermal and catalytic cracking as well as coke formation using decay functions, one as a function of time and the other as a function of coke concentration on catalyst [194]. A common problem with these models is accounting for the immiscibility of the different polyolefins in waste mixtures. Kruse et. al. presented a binary model for a PS-PP mixture, consisting of reactions [195]. This binary model was developed by the combination of two original models, each describing PS and PP degradation respectively. By premixing the two polymers and slightly adjusting the PS to PP diffusion parameter, the model could portray binary interactions in PS- PP pyrolysis, including the enhancement of PP degradation by PS radicals formed during decomposition [195]. On the other hand, Faravelli et. al. presented a kinetic model that shows that accounting for binary interactions causes a deviation from experimental data, arguing that binary interactions can only be observed at high macro-mixing levels [151]. TG is a good kinetic estimation tool; however, kinetic information obtained by TG is characterized by errors due to certain restrictions such as heat and mass transfer limitations. Similarly, mechanistic modeling requires a lot of assumptions and is computationally 39

56 intensive; therefore, whenever possible, it may be best that kinetic data be estimated directly from experimental data. As shown, there has been a lot of work done in the pyrolysis of polyolefins. However, the investigations of polyolefin pyrolysis with FCCs are scarce in literature. Furthermore, an estimation of kinetic parameters for the catalytic pyrolysis of polyolefins by TG methods has been scarcely done in literature. Therefore, it is the aim of this study to contribute to the area by investigating the kinectics of polyolefin pyrolysis using FCCs. 40

57 CHAPTER 3 EXPERIMENTAL METHODS This chapter reviews the methods by which the polymer samples were prepared. First, the polymer and catalyst materials are described. Additionally, the various catalyst characterization methods employed are briefly outlined. Next, the various methods by which thermogravimetry (TG) data is used to in the evaluation of the catalysts on polypropylene pyrolysis are explained. 3.1 Polymer Materials Polypropylene (PP), P4CZ-027, was supplied by Huntsman Corporation from Utah, while high density Polyethylene (PE), DGDA-6944 NT, was supplied by Union Carbide from New Jersey. Polystyrene (PS), MC3600, was supplied by Chevron Phillips Chemical Company. In this study, shredded post consumer carpet (PP-PCC) consisting of PP face yarns was supplied by Wellman, Inc. (Post consumer carpet with nylon face yarns are being evaluated concurrently by Bryson [196].) The composition of this material will be discussed below Composition of Polypropylene Post Consumer Carpet (PP-PCC) Carpet consists of a face yarn with primary and secondary backing fabric/material glued together by a latex rubber product. The face yarns are typically made of nylon, polypropylene, acrylic, or polyester. The primary and secondary backing fabrics are usually PP. The latex adhesive is usually styrene-butadiene rubber (SBR)[197, 198]. Calcium Carbonate (CaCO 3 ) is typically added as filler in the latex adhesive as shown in Figure

58 Figure 3.1 Structure of tufted carpet [198] It was estimated that the shredded carpet material has a weight composition of 84% PP face fiber including backing material and 9% nylon 6 fibers [199]. PP-PCC is also expected to consist of contaminants such as dirt as is commonly observed with post consumer carpets. This is illustrated in Figure 3.2 below. Figure 3.2 Composition of post consumer carpet [198] Preparation of PP-PCC Pellets The shredded PP- PCC was converted into pellets using an A-Class type 55 VSP model of a Next Generation Recycling (NGR) repelletizing system. The NGR unit consists of a single screw extruder and a shredder incorporated in the feed section of the extruder. The shredded carpet is melted in the extruder barrel. This melt is filtered through a 20-mesh screen, 42

59 extruded as long continuous strands which are cooled in a cold water bath. These strands are collected and cut up into cylindrical shaped pellets using a Reiter pelletizer [199]. 3.2 Catalyst Materials ZSM-5 ZSM-5 was synthesized hydrothermally from clear gel solutions following the general procedure outlined for the synthesis of nanocrystalline ZSM-5 by Van Grieken et. al.[200]. In preparation of the gel solutions, measured amounts of water, sodium hydroxide (NaOH), tetrapropylammonium hydroxide (TPAOH, 40wt% aqueous solution), tetraethyl orthosilicate (silicon source) and aluminium nitrate (aluminum source) were mixed using the following composition: 9TPAOH : 0.16NaOH : Al :25Si :495H20 The hydrothermal treatment was conducted in an autoclave equipped with a rotating sample holder for continuous stirring at 165 C for 116 hours. The resulting solution was a whitish colloidal mixture of crystals and solvent. After three cycles of washing with water and centrifugation, the sample was dried at 120 C for one hour. Subsequently, the dried sample was calcined at 550 C under airflow for 5 hours. The methods of characterization for the synthesized ZSM-5 sample are discussed below Xray Diffraction The calcined ZSM-5 crystals were characterized by X-ray diffraction (XRD) using a Cu K alpha target ( A). XRD patterns were collected between 2θ angles of 0 and 55. The XRD pattern is shown in Figure

60 Figure 3.3 XRD patterns of synthesized ZSM-5 The characteristic peaks between 5 and 10 2θ angles are indicative of a highly crystalline solid and showed a typical MFI-zeolite framework, which is characteristic of ZSM-5. Furthermore, the sharp narrow peaks indicate a high periodicity of the crystalline domains [201] Solid State Magic Angle Spinning (MAS) NMR 29 Si and 27 Al solid state MAS NMR spectra were obtained using a 7.4 T wide bore magnet ( 1 H frequency at 300 MHz) with 7mm outer diameter rotors MAS probe spinning at 5 khz and 6.5 khz, respectively. For 29 Si, each single pulse spectrum was acquired for 128 scans 44

61 with a 60 seconds pulse repetition delay and calibration with respect to DSS at 0 ppm. For the 27 Al spectra, signal averaging of 1024 scans with a 3 seconds pulse delay and calibration with respect to Al 3+ ions at 0 ppm was used. The results for both spectra are shown in Figures 3.4 (a) and 3.4 (b). The signal at in the 27 Al spectrum is characteristic of tetrahedral coordination of Al atoms, whereas the lower signals are typical of calcined ZSM-5 samples. 45

62 (a) (b) Figure Si (a) and 27 Al (b) solid state MAS NMR spectrum of calcinated ZSM-5 samples. 46

63 Scanning Electron Microscopy (SEM) SEM images of ZSM-5 were obtained using a Hitachi S800 field emission gun (FEG) SEM. Shortly before a SEM image was taken, the sample was coated with gold. The image of the ZSM-5 crystals shown in Figure 3.5 was taken at a 512 x 512 resolution. Figure 3.5 SEM image of ZSM-5 at 1000X Furthermore, SEM was combined with energy dispersive X-ray analysis (EDX) in the detection of the elemental composition of ZSM-5 as reported in Figure

64 Figure 3.6 EDX bulk plot for synthesized ZSM-5 The results report a Si/Al ratio of 30.4 which is 21% off the original gel composition (Si/Al = 25). This could be attributed to errors in the SEM-EDX analysis, which may not be the paramount method for Si/AL determination. Additionally, errors may also arise from the loss of material during zeolite synthesis Nitrogen Adsorption Isotherms Nitrogen adsorption isotherms were obtained on a Micromeritics ASAP 2010 using approximately 0.23g of sample. Prior to nitrogen adsorption, the sample was vacuum degassed for approximately 1 hour at 120 C. The results obtained are shown in Table 3.1 below. 48

65 Table 3.1 Nitrogen adsorption isotherm report on ZSM-5 sample BET surface area, m 2 /g ± 5.07 Micropore area, m 2 /g Micropore volume cm 3 /g 0.11 Although, a BET surface area was reported here, it should be noted that the BET approximation does not work well with microporous materials. This is because the BET equation is based upon the formation of multiple layers, which are not easily formed in microporous materials [202] Fluid Catalytic Cracking (FCC) Catalysts The FCC catalysts employed in this study are believed to be composed of a Y-zeolite. Used (or equilibrium) FCC catalysts were supplied by Shell and Albemarle and are abbreviated as S-ECAT and A-ECAT, respectively. In addition, fresh FCC catalyst samples having two different particles sizes were also supplied by Albemarle and are labeled Fresh CAT and Fresh Fines for the larger and smaller sizes respectively. The physical properties of these catalysts are displayed in Table

66 Table 3.2 Physical properties of Albemarle catalysts as reported by supplier Fresh CAT Fresh Fines A-ECAT Zeolite SA (m 2 /gm) Total SA (m 2 /gm) Particle size distribution (microns) Nickel (ppmw) 3488 Vanadium (ppmw) 3938 Antimony (ppmw) 1312 It is not clear how the particle size distribution is estimated; however, when such values are reported, they typically represent a peak particle size within a volume based distribution of particle sizes. Additional characterizations were carried out on the FCC catalysts and these are discussed Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray (EDX) SEM images of S-ECAT and the Albemarle fresh FCC catalysts are shown in Figures 3.7(a), 3.7(b) and 3.7(c), respectively. These were obtained at magnifications of 200X, 500X and 500X, respectively. 50

67 Figure 3.7 (a) SEM image of S-ECAT taken at 200X Figure 3.7 (b) SEM image of Fresh FCC Fines taken at 500X 51

68 Figure 3.7 (c) SEM image of Fresh CAT (regular) taken at 500 X From the SEM images, it may be deduced that S-ECAT particles sizes range within µm, whereas the Fresh CAT and Fresh FCC Fines measure µm and µm, respectively. In further characterizing S-ECAT, SEM was combined with EDX in the detection of the zeolitic and metal compositions as reported in Figure

69 Figure 3.8 EDX bulk plot for S-ECAT The analysis reports a Si/Al ratio of approximately 1.2 for S-ECAT. This value is very close to that observed in a Y-zeolite (approximately 2) Particle Size Distribution (PSD) by SEM SEM images of the fresh FCC catalysts were taken and loaded on a Vashall image analyzer where the particle size determinations were performed by analyzing 111 and 102 particles for the Fines and regular sized FCC catalysts, respectively. A plot of the mean diameters observed for the individual particles is shown in Figure 3.9. The analysis yielded average PSD diameters for the Fines and the regular size particles are 18 ± 11 and 21 ±16 microns, respectively. This is different from the average diameters reported by the supplier (shown in Table 3.2). 53

70 fines regular Mean diamter (microns) Particle # Figure 3.9 PSD analyses by SEM for Fresh FCC catalysts Figure 3.9 clearly shows much variation of particle mean diameters within both samples Calcium Carbonate (CaCO 3 ) CaCO 3 was supplied by Fisher Scientific (CAS: ). CaCO 3 is typically used as filler in the latex adhesive in carpets. The catalytic effect of CaCO 3 in the TG pyrolysis of PP-PCC was assessed and will be discussed in the next chapter. 3.3 Thermogravimetry TG involves monitoring the weight loss of the sample as a controlled function of temperature. An important application of TG in the study of polymers is the measurement of thermal stability. In this study, this analysis serves primarily as an assessment tool in the 54

71 screening of various potential catalysts for polyolefin pyrolysis. An assessment of the catalyst performance prior to use in a reactor reduces costs. The thermogravimetry analysis (TGA) equipment used in this study is a Seiko TG/DTA 320 running on the EXSTAR6000 software package. An inert atmosphere of high purity nitrogen gas, flowing at 300mL/min was used. The balance can hold a maximum of 15mg; therefore, all sample amounts used in this study averaged approximately 10.2 mg (min: 9.8mg and max: 12.5 mg) and a measurement range of C was used for all runs. This range was chosen to ensure that all possible decomposition steps are identified for each polymer sample. (Higher temperatures were not attempted because aluminum pans were used and these start to anneal above 600 C.) Furthermore, this particular TG model allows for the variation of parameters such as the heating rate. In addition, to the mass at each temperature change, the data output also includes time and rate of weight loss (DTG) at each temperature step. The heating rate feature and time data are particularly important in the kinetic study of the pyrolysis reactions, which will be discussed in Section TG Sample Preparation In the TG study of the catalytic pyrolysis of PE by MCM41, USY and ZSM-5, Marcilla et al. prepared their samples by mixing dried proportions of polymer and catalyst ( dry mixing ) [25]. In their analysis, they found the amount of catalyst to be different for runs at the same concentration. This makes it difficult to assess the exact performance of the catalysts, especially when the decomposition curves are very similar [25]. This is apparently attributable to the method of sample preparation used. Alternatively, one could mix in the 55

72 catalyst into the polymer melt. In this study, the polymer is melted in a Haake Rheomix 600. The Rheomix 600 has three zones that are heated to set temperatures and mixing is achieved by rotation of the two sigma blades as shown in Figure Figure 3.10 Schematic of mixer cross-section showing sigma blade placement in the mixing zone The operating temperature is set very close to that of the polymer s melting temperature range, T m. The operating temperatures used to obtain the various mixtures of catalysts and polymer (PP, PS and HDPE) are shown in Table 3.3. Also highlighted in Table 3.3 are the steps taken in obtaining TG samples by melt-mixing. 56